Crate netform_ir 

lossless intermediate representation (IR) for network device configuration text.

this crate provides:

• a tree model (Document, Node, LineNode, BlockNode)
• a conservative parser (parse_generic, parse_with_dialect)
• a lossless renderer (Document::render)

the parser is intentionally conservative:

• it only uses indentation as a structural cue
• unknown patterns are preserved as regular lines
• no input lines are dropped

Example

use netform_ir::parse_generic;

let input = "interface Ethernet1\n  description uplink\n";
let doc = parse_generic(input);
assert_eq!(doc.render(), input);

Modules

detect

score-based dialect auto-detection from configuration text.

Structs

BlockNode

structured block node with a header line and nested children.

Document

lossless parsed document backed by an arena and root node list.

DocumentMetadata

document metadata attached during parsing.

GenericDialect

conservative default dialect for vendor-agnostic parsing.

IosKeyHintConfig

configuration for ios_family_key_hint, parameterizing the constructs that differ across IOS-family dialects while sharing the common structure.

IosLikeDialect

parameterized dialect for IOS-like configuration text (EOS, IOS XE, NX-OS, …).

LineNode

leaf node preserving original raw text and parse metadata for one line.

NodeId

stable arena identifier for a node in a Document.

ParseFinding

parser-level uncertainty note attached to a source span.

ParsedLineParts

minimal tokenized representation of a content line.

Path

location path used by diffs and diagnostics (root_index, then child indices).

Span

source span pointing to a single line and byte range in the original input.

Enums

DialectHint

declared parser dialect used for this document.

LiteralTerminator

what closes a multi-line literal region opened by Dialect::literal_region.

Node

arena node variant.

TriviaKind

lightweight classification used by parser, normalization, and diff views.

Traits

Dialect

dialect extension point for trivia classification and line tokenization.

Functions

classify_ios_like_trivia

classify trivia for IOS-like dialects (EOS, IOS XE).

classify_trivia_with_prefixes

classify a line as blank, comment, or content based on the given comment prefixes.

common_key_hint

derive a stable identity key for constructs shared across all IOS-like dialects (EOS, IOS XE, NX-OS).

count_indent

count the visual indentation width of a line, treating tabs as 4 spaces.

ends_inside_quoted_value

report whether raw ends with a double-quoted value still open.

ios_family_key_hint

derive a key hint for constructs shared across IOS-family dialects.

ios_like_literal_region

recognize an IOS-family banner <type> [delimiter] line and report what closes the free-text body that follows it.

parse_generic

parse input using the built-in generic dialect.

parse_interface

parse an interface name into (canonical_type, id) using the given type prefix table.

parse_ios_like_parts

tokenize a content line for IOS-like dialects (EOS, IOS XE).

parse_with_dialect

parse input into a lossless IR using the given dialect implementation.

tokenize

tokenize a raw configuration line, splitting on whitespace while preserving quoted strings. characters in punctuation are emitted as their own tokens (flushing any accumulated word first), matching the Junos-style brace/semicolon handling.